[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Journalling filesystems

From: Marcus Brinkmann
Subject: Re: Journalling filesystems
Date: Wed, 16 Jun 2004 17:52:13 +0200
User-agent: Wanderlust/2.10.1 (Watching The Wheels) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.3 (i386-pc-linux-gnu) MULE/5.0 (SAKAKI)

At Wed, 16 Jun 2004 16:46:41 +0200,
Bas Wijnen wrote:
> [1  <multipart/signed (7bit)>]
> [1.1  <text/plain; us-ascii (quoted-printable)>]
> On Tue, Jun 15, 2004 at 11:18:35PM +0200, Marco Gerards wrote:
> > This is a really important feature for a server os to have, IMHO.
> Not just for a server OS.  Any OS would get rock solid from it.  The problem
> that a crashing computer kills your installation is only partly gone with
> journalling as in ext3.  With transactions there is no way a crash can destroy
> the system, not even if it happens during an upgrade of critical components.
> Unless the crash is in the hardware, notably the hard disk, of course.

Of course, this is not only an OS issue, but also an application
issue.  As was correctly pointed out, the application needs to provide
hints when a transaction starts and ends.

This is very powerful, but also has drawbacks (performance, apps need
to change).  Anyway, things like this is what the Hurd design wants to
make easy to add in userspace, so go for it.

However, because it is not intrinsically OS-specific, it seems worth
to me to make this a broad effort for Unix-like systems in general

I also have the feeling that there are other designs that can provide
an alternative or supplement to this (mmh, persistence?
check-pointing?  Versioning?)

In any case, this whole issue seems to me to be big enough for a dozen
or so independent projects for themselves.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]